Uber Data를 활용한 분석

Python
Code
h3
Author

Jinwon Lee

Published

May 29, 2023

h3 라이브러리를 활용한 Uber Data 분석

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from h3 import h3
from collections import Counter
import pydeck as pdk

# 한글 폰트 사용을 위해서 세팅
from matplotlib import font_manager, rc, rcParams
font_path = "C:/Windows/Fonts/malgun.ttf"
font = font_manager.FontProperties(fname = font_path).get_name()
rc('font', family = font)

# 마이너스 깨짐 현상 해결
rcParams['axes.unicode_minus'] = False
H3_HEX_DATA = "https://raw.githubusercontent.com/visgl/deck.gl-data/master/website/sf.h3cells.json"

df = pd.read_json(H3_HEX_DATA)
df.head()
hex count
0 88283082b9fffff 96
1 8828308281fffff 534
2 88283082d7fffff 36
3 88283082c1fffff 297
4 88283082a9fffff 147 
def h3_to_lat_lon(h3_address):
    lat_lon = h3.h3_to_geo(h3_address)
    return lat_lon

def lat_lon_to_h3(lat, lon):
    h3_address = h3.geo_to_h3(lat, lon, resolution = 6) # resolution : 해상도
    return h3_address
df['lat_lon'] = df['hex'].apply(h3_to_lat_lon)
df.head()
Date/Time Lat Lon Base Datetime Weekday Hour hex lat_lon
0 4/1/2014 0:11:00 40.7690 -73.9549 B02512 2014-04-01 00:11:00 Tuesday 0 862a100d7ffffff (40.74274905936817, -73.95762841287764)
1 4/1/2014 0:17:00 40.7267 -74.0345 B02512 2014-04-01 00:17:00 Tuesday 0 862a1072fffffff (40.7041679083504, -74.01600847979245)
2 4/1/2014 0:21:00 40.7316 -73.9873 B02512 2014-04-01 00:21:00 Tuesday 0 862a100d7ffffff (40.74274905936817, -73.95762841287764)
3 4/1/2014 0:28:00 40.7588 -73.9776 B02512 2014-04-01 00:28:00 Tuesday 0 862a100d7ffffff (40.74274905936817, -73.95762841287764)
4 4/1/2014 0:33:00 40.7594 -73.9722 B02512 2014-04-01 00:33:00 Tuesday 0 862a100d7ffffff (40.74274905936817, -73.95762841287764) 
# Define a layer to display on a map
layer = pdk.Layer(
    "H3HexagonLayer",
    df,
    pickable=True,
    stroked=True,
    filled=True,
    extruded=False,
    get_hexagon="hex",
    get_fill_color="[255 - count, 255, count]",
    get_line_color=[255, 255, 255],
    line_width_min_pixels=2,
)

# Set the viewport location
view_state = pdk.ViewState(latitude=37.7749295, longitude=-122.4194155, zoom=14, bearing=0, pitch=30)

# Render
r = pdk.Deck(layers=[layer], initial_view_state=view_state, tooltip={"text": "Count: {count}"})
r.show()
# 데이터 로드
df = pd.read_csv('https://github.com/fivethirtyeight/uber-tlc-foil-response/raw/master/uber-trip-data/uber-raw-data-apr14.csv')

# 데이터셋 확인
print(df.head())
          Date/Time      Lat      Lon    Base
0  4/1/2014 0:11:00  40.7690 -73.9549  B02512
1  4/1/2014 0:17:00  40.7267 -74.0345  B02512
2  4/1/2014 0:21:00  40.7316 -73.9873  B02512
3  4/1/2014 0:28:00  40.7588 -73.9776  B02512
4  4/1/2014 0:33:00  40.7594 -73.9722  B02512
df.describe()
Lat Lon
count 564516.000000 564516.000000
mean 40.740005 -73.976817
std 0.036083 0.050426
min 40.072900 -74.773300
25% 40.722500 -73.997700
50% 40.742500 -73.984800
75% 40.760700 -73.970000
max 42.116600 -72.066600

1. 데이터의 기본적인 indexing 및 slicing

# 처음 5개의 행을 출력
df.head()
Date/Time Lat Lon Base
0 4/1/2014 0:11:00 40.7690 -73.9549 B02512
1 4/1/2014 0:17:00 40.7267 -74.0345 B02512
2 4/1/2014 0:21:00 40.7316 -73.9873 B02512
3 4/1/2014 0:28:00 40.7588 -73.9776 B02512
4 4/1/2014 0:33:00 40.7594 -73.9722 B02512 
# 'Lat' 열의 값만 출력
df['Lat']
0         40.7690
1         40.7267
2         40.7316
3         40.7588
4         40.7594
           ...   
564511    40.7640
564512    40.7629
564513    40.7443
564514    40.6756
564515    40.6880
Name: Lat, Length: 564516, dtype: float64
# 10행부터 20행까지 출력
df.iloc[10:21]
Date/Time Lat Lon Base
10 4/1/2014 1:19:00 40.7256 -73.9869 B02512
11 4/1/2014 1:48:00 40.7591 -73.9684 B02512
12 4/1/2014 1:49:00 40.7271 -73.9803 B02512
13 4/1/2014 2:11:00 40.6463 -73.7896 B02512
14 4/1/2014 2:25:00 40.7564 -73.9167 B02512
15 4/1/2014 2:31:00 40.7666 -73.9531 B02512
16 4/1/2014 2:43:00 40.7580 -73.9761 B02512
17 4/1/2014 3:22:00 40.7238 -73.9821 B02512
18 4/1/2014 3:35:00 40.7531 -74.0039 B02512
19 4/1/2014 3:35:00 40.7389 -74.0393 B02512
20 4/1/2014 3:41:00 40.7619 -73.9715 B02512 
# Base 갯수 확인
df['Base'].value_counts()
Base
B02682    227808
B02598    183263
B02617    108001
B02512     35536
B02764      9908
Name: count, dtype: int64

2. 결측치에 대한 처리

# 결측치 개수 확인
df.isnull().sum()
Date/Time    0
Lat          0
Lon          0
Base         0
dtype: int64

3. 기초적인 통계 추출 및 분석

# 데이터 요약
print(df.describe())

# 'Lat' 열의 평균
print(df['Lat'].mean())

# 'Lon' 열의 중앙값
print(df['Lon'].median())

# 'Base' 열에서 각 값의 빈도수 출력
print(df['Base'].value_counts())
                 Lat            Lon
count  564516.000000  564516.000000
mean       40.740005     -73.976817
std         0.036083       0.050426
min        40.072900     -74.773300
25%        40.722500     -73.997700
50%        40.742500     -73.984800
75%        40.760700     -73.970000
max        42.116600     -72.066600
40.74000520746974
-73.9848
Base
B02682    227808
B02598    183263
B02617    108001
B02512     35536
B02764      9908
Name: count, dtype: int64

4. 데이터에 대한 질문을 던져보고 해답을 찾기

4.1 월별/요일별 Uber 이용량 비교

  • 가장 Uber 이용량이 많은 요일은 언제일까요?
df['Datetime'] = pd.to_datetime(df['Date/Time'])
df['Weekday'] = df['Datetime'].dt.day_name()
weekday_count = df['Weekday'].value_counts().to_frame().reset_index()
display(weekday_count)
print(f"Uber 이용량이 가장 많은 요일 : {weekday_count.loc[weekday_count['count'].idxmax(), 'Weekday']}")
Weekday count
0 Wednesday 108631
1 Tuesday 91185
2 Friday 90303
3 Thursday 85067
4 Saturday 77218
5 Monday 60861
6 Sunday 51251 
Uber 이용량이 가장 많은 요일 : Wednesday
fig, ax = plt.subplots(figsize = (8, 6))
sns.barplot(x = 'Weekday', y = 'count', ax = ax, data = weekday_count, palette = 'Set2')

plt.title('요일별 Uber 이용량')
plt.show()

4.2 시간대별 이용량 비교

  • 가장 이용량이 많은 시간대는 언제일까요?
df['Hour'] = df['Datetime'].dt.hour
hour_count = df['Hour'].value_counts().to_frame().reset_index()
display(hour_count.head())
print(f"Uber 이용량이 가장 많은 시간대 : {hour_count.loc[hour_count['count'].idxmax(), 'Hour']}")
Hour count
0 17 45475
1 18 43003
2 16 42003
3 19 38923
4 21 36964 
Uber 이용량이 가장 많은 시간대 : 17
fig, ax = plt.subplots(figsize = (8, 6))
sns.barplot(x = 'Hour', y = 'count', ax = ax, data = hour_count, color = '#8BB0F5')# palette = 'viridis')

plt.title('시간대별 Uber 이용량')
plt.show()

4.3 Uber 수요에 대한 공간적인 시각화

  • 가장 수요가 많은 지역은 어디일까요?
  • 시간대/지역별로 수요의 변화를 나타내보고, 시사점을 도출해 봅시다
df['hex'] = df.apply(lambda row : h3.geo_to_h3(row['Lat'], row['Lon'], 6), axis = 1)
df.head()
Date/Time Lat Lon Base Datetime Weekday Hour hex
0 4/1/2014 0:11:00 40.7690 -73.9549 B02512 2014-04-01 00:11:00 Tuesday 0 862a100d7ffffff
1 4/1/2014 0:17:00 40.7267 -74.0345 B02512 2014-04-01 00:17:00 Tuesday 0 862a1072fffffff
2 4/1/2014 0:21:00 40.7316 -73.9873 B02512 2014-04-01 00:21:00 Tuesday 0 862a100d7ffffff
3 4/1/2014 0:28:00 40.7588 -73.9776 B02512 2014-04-01 00:28:00 Tuesday 0 862a100d7ffffff
4 4/1/2014 0:33:00 40.7594 -73.9722 B02512 2014-04-01 00:33:00 Tuesday 0 862a100d7ffffff 
h3_counts = Counter(df['hex']) # 각 셀에 포함된 데이터 개수 계산
df_h3_counts = pd.DataFrame.from_dict(h3_counts, orient = 'index').reset_index()
df_h3_counts.columns = ['h3', 'count']

df_h3_counts['lat'] = df_h3_counts['h3'].apply(lambda x : h3.h3_to_geo(x)[0])
df_h3_counts['lon'] = df_h3_counts['h3'].apply(lambda x : h3.h3_to_geo(x)[1])

df_h3_counts.head()
h3 count lat lon
0 862a100d7ffffff 247294 40.742749 -73.957628
1 862a1072fffffff 125962 40.704168 -74.016008
2 862a10727ffffff 50036 40.759208 -74.033187
3 862a103b7ffffff 11639 40.654664 -73.789791
4 862a100f7ffffff 10999 40.781318 -73.899155 
# Define a layer to display on a map
layer = pdk.Layer(
    'H3HexagonLayer',
    df_h3_counts,
    pickable = True,
    stroked = True,
    filled = True,
    extruded = False,
    get_hexagon = 'h3',
    get_fill_color = '[255 - count, 255, count]',
    get_line_color = [255, 255, 255],
    line_width_min_pixels = 2,
)

# Set the viewport location
view_state = pdk.ViewState(latitude = 40.7425, longitude = -73.9848,
                           zoom = 8, bearing = 0, pitch = 30)

# Render
r = pdk.Deck(layers = [layer], initial_view_state = view_state, tooltip = {'text' : 'Count: {count}'})
r.show()